Prof. Dr. Alexander Zipf*
Chair of Cartography
*Department of Geography
University of Bonn, Germany
By combining the growing amount of user generated geo-content, 3D visualisations and open standards, new vistas can be opened up for mobile navigation services. The possibilities are there, now it is time to integrate these resources into innovative mobile products. This convergence of 3D, mobile and Web2.0 approaches is the first sign that the dream of Ubiquitous GI Services (UbiGIS)is finally materialising.
When considering "openness" in the context of mobile navigation services, there are a few technologies, standards and institutions one cannot ignore. Specifically the Open Geospatial Consortium (OGC) standards for the interoperable management, processing and visualisation of spatial data has to be covered. For these standards we nowadays have very usable and stable open source libraries as well as commercial products. By combining and enhancing these, innovative new concepts and solutions can be achieved also in the domain of Location Based Services (LBS).
STANDARDS FOR MOBILE NAVIGATION
The most relevant OGC standard with respect to LBS is the OpenGIS Location Services specification, a series of implementation specifications initially covering five core services:
- Directory Service
- Gateway Service
- Location Utility Service
- Presentation Service
- Route Service
While the Utility Service (GeoCoder and Reverse GeoCoder) is moving to the OGC mass market initiative, only recently a sixth service – the Tracking Service – has been defined, that opens the way for a wealth of new interesting LBS applications. A further draft specification – the Navigation Service – has been in discussion since long. In the forthcoming version 1.3 of the OpenLS specification it is planned to add it as an enhancement of the Route Service. Most of these OpenLS core services have been implemented by us and shall be made open source soon. Several enhanced versions of the route service have been applied as spin-offs in a number of scenarios:
- Emergency Route Service (ERS) – The ERS is a special OpenLS Route Service, that considers specified areas to be avoided (e.g. flooded or blocked roads, landslides, poisoned areas) while calculating the requested route. (Neis et al. 2007).
- Accessibility Analysis Service (AAS) – This is a service that calculates a polygon around a certain start location (e.g.: city, point of interest, address). That polygon represents the area that that is reachable from the start location within a certain time or a defined distance. (Neis & Zipf 2007).
- Further the Route Service 3D (RS3D) and Route Service with Landmarks and Focus Maps will be explained later, see also Neis et al (2007). It combines both approaches, by integrating 3D Landmarks in a 3D scene and delivered from an OGC W3DS instead of a WMS.
FREE, COLLABORATIVELY COLLECTED DATA With the recent boom in Web 2.0 and the emergence of user generated geodata, some noticeable projects have showed up on the web. Well known, but especially noteworthy is Open- StreetMap.org, a project that provides geodata collected by their user via GPS. This data is not only accessible through their web interface – you can also download their street network and thus employ it for your own application. The amount of free data provided by OpenStreetMap is already huge and continuously growing.
It is time to use this large quantity of free information for more than just simple web mapping. The project OpenRouteService.org provides free routing based on OpenStreetMap data through our implementation of the OpenLS Route Service (Neis & Zipf 2008). It also will be made available as Open Source soon. Through the integration with the Geocoder functionality of the Location Utility Service also an interoperable solution for searching for addresses will be offered. An interesting feature that stems from our work on disaster management is the possibility to avoid areas that are not passable. Originally those areas were defined by the emergency service staff, but we are currently implementing the functionality that lets the user interactively digitise areas to be excluded from the routing algorithm.
Thus users are able integrate their local knowledge about construction sites and suchlike into their routes. In a next step, this can be developed into Web 2.0 application of its own, by enabling users to share their knowledge about areas best avoided with others. Figure 1 shows how one bridge has been declared impassable (red polygon) and the route now uses the northern bridge.
INTEROPERABLE SDI – A FUNDAMENTAL REQUIREMENT FOR LBS
Interoperability achieved through open specifications is of particular importance in the mobile world of LBS, where users roam through different regions with different providers using an extremely heterogeneous set of devices and clients. In order to keep everything as open and interoperable as possible, we use the specifications of the OGC. The most relevant OGC web services next to the OpenLS initiative for our work are:
- Web Catalogue Service
- Web Map Service
- Web Feature Service
- Web 3D Service
- Web Processing Service
These are used to set up domain specific SDIs, e.g. in the area of disaster management (see www.okgis.de for a project on open source and open SDIbased disaster management) or on 3DSDI in projects like www.gdi-3d.de (Schilling et al 2006, 2007). A similar approach is taken in our current project on mobile navigation with 3D city models (www.MoNa3D.de).
Some approaches and results from the latter will be presented in the following sections. The conceptual architecture of GDI-3D.de is presented in figure 2.
ADDING THE THIRD DIMENSION
In those 3D projects we use the Web 3D Service (W3DS), another pending OGC specification. It offers the possibility to generate interactive 3D scenes of city models and digital elevation models (DEM) from distributed data sources in various 3D formats. In order to realise a functional 3D routing service it is not enough to intersect the geometry of a route with a DEM, but we need also to consider bridges, tunnels, over- and underpasses etc. This has been successfully realised in a web-based environment. (see figure 3).
Any appropriate client can now access this SDI and get a route calculated as well as visualised in a 3D environment. In our own client, we have additional features like a flight along the route or highlighted decision points for an optimised result (see figure 4).
Extensive research has shown that the navigation experience can be enhanced
Fig. 1: Extended Route Version of Open- RouteService.org: user defined areas are avoided
Fig. 2: OGC Services in the Heidelberg 3D-SDI (www.gdi-3d.de)
significantly by including landmarks into the route instruction. We currently are developing another extension to the OpenLS Route Service, which will support the usage of fixed Points of Interest as landmarks as well as more dynamic approaches. For greater effectiveness, the visibility as well as inherent properties of the objects shall be considered during the selection of suitable landmarks. These functionalities will be realised using the new OGC Web Processing Service (WPS).
3D VISUALISATION RULES FOR NAVIGATION
Now we need to visualise the 3D scenes in an appropriate and interoperable manner. Typically, within GIS there is a clean division between the raw geodata and the visualisation properties. This is an advantage because the same data can be used and displayed in multiple ways according to the specific needs. This division should also apply to 3D city models and mobile applications. In order to allow a clean cut between raw data and visualisation, a separate open format for the visualisation rules needs to be defined. This is already done for 2D (web) maps through the OGC Styled Layer Descriptor (SLD) specification – or more precisely the newer OGC Symbology Encoding specification (SE). This offers many opportunities. Apart from allowing a client application or end user to define the style of a map, more importantly it makes it possible to integrate diverse data sources into one WMS map and to style them consistently within that map. For the same reasons it would be very beneficial if this mechanism could also be applied to 3D data representing DEMs, 3D landscape and city models.
Therefore we have developed (Neubauer & Zipf 2006) an extension to the current OGC Symbology Encoding specification, that is implemented into our W3DS (Neubauer et al. 2007). This can be used to define the appearance of the 3D scene directly from the client side, such as it is the case with 2D WMS. In particular this allows further to use the OGC Filter Encoding functionalities for thematic filtering, to select 3D objects like buildings based on attribute values. The selected buildings than receive their specific visualisation properties through the SE too. It is also possible to include external 3D graphics (e.g.in form of VRML models) for point-objects. This allows to change the representation of objects on the fly. One example for that is the 3D representation of landmarks can be charged dynamically. This can, for example, be used to provide different user groups with different visualisations or to adapt the visualisation according to traffic modalities or order changes of the context. An example of this are 3D focus maps, as shown in figure 5. This is an extension of the original 2D focus maps (Zipf & Richter 2002). When calculating routes or maps for navigation, it includes and distinguishes automatically relevant objects in order to assist the user to focus on the most significant parts of the map.
The combination of mobile clients and spatial data infrastructures should provide a means to direct to clients to the right data providing service depending on its location. This can be handled by a Web Catalogue Service where all the necessary data sources are listed and the client gets directed to the appropriate server.
Mobile navigation systems also need to synchronise the time and/or place a route instruction has to be delivered. To achieve this, we are currently planning to advance our Route Service towards a Navigation Service, so mobile navigation can be fully supported.
When working with large scale, detailed datasets in combination with mobile applications, the data transfer to the client always constitutes a bot-
Fig. 3: Dynamic OpenLS 3D routing supporting bridges, underpasses etc. (Schlling et al. 2008)
Fig. 4: Visualising a route in a 3D SDI environment
tleneck. A large part of the data that needs to be transferred consists of textures for the 3D models, slowing down the performance decidedly – but displaying only mono-colored buildings is not satisfactory either. In order to solve this dilemma, our partners at the University of Technology Stuttgart are developing a smart algorithm to decompose building textures into essential parts. Then these parts are transferred to the client along with a so called pulse function, determining how they are to be assembled so as to form a synthetic texture that appears realistic. (Coors & Zipf 2007).
PERSONALISED AND CONTEXT AWARE LBS
A further set of new specifications (or drafts) we are working with deal with the integration of sensor data: the so called Sensor Web. Dynamic data about a lot of information on the environment (weather, floodings, etc.) or the traffic situation (traffic jams, accidents, construction works etc.) or even surveillance cameras can be sensed and distributed in near real time now. The need of integration of these extremely heterogeneous data sources through open standards is becoming more pressing. The variety of new sensor data will allow us to develop adaptive services that offer personalised and situation aware information.
This has been proposed for a few years in particular in the area of mobile maps (e.g. for tourism, Zipf 2002), but now we see that more and more dynamic data sources are becoming available that allow to realise this in a wider context to the point of Ubiquitous Computing (UbiComp). A term that may be defined as: Pervasive services based on Ubi- Comp technology and devices, supporting context-dependent (i.e. adaptive) interaction, realised by information and functions of geographic information services based on interoperable SDI (Reuter & Zipf 2008). It has been pointed out earlier, that the adaptivity of GI services to context can be seen as one of the next steps for GIScience research in order to achieve more intuitively usable GI systems (Zipf 1998). A few ideas of which adaptive services might be suitable within the context of GI services:
- adaptation of the visual presentation of the contents offered – both textual and graphical (pictures, maps, video, VR models);
- adaptation of route planning (by individual weighting and restrictions);
- adaptation of queries (combined location- and interest-based tips;
- adaptation of the offered contents (e.g. concerning details, topic).
It has been shown that the extensive suite of OGC specifications (or draft specifications) available today is quite rich and enables the development of interesting LBS or mobile and webbased GI applications in general. Of course there are open issues, in particular when it comes to more fine grained visualisation rules for 3D maps as well as thematic maps, where the current SLD/SE approach is too limited and needs extensions (Dietze & Zipf 2007).
While processing and analysis of spatial data can now be integrated into a service chain of OGC services through the Web Processing Service (WPS) specification in general, there is still a lot of research about how to deal with WPS functionalities in detail (see Stollberg & Zipf 2007,2008, Goebel & Zipf 2008). These service chains can be also realised in the area of navigation and 3D-GIS (Zipf et al. 2007), but earlier we have identified some technical challenges for dynamic web service orchestration based on the Business Process Execution Language (BPEL) (Weiser & Zipf 2007) in a different, but comparable application domain. These problems should disappear soon, as technologies mature and OGC services will support further industry-standards such as WSDL (Web Service Description Language).
It is our belief and motivation, that in combination the introduced technologies can lead to more pervasive GI services and so can contribute to the development of the next generation of mobile navigation systems. Maybe they even show the way to something for which the term Ubiquitous Geographic Information = UbiGIS could be applicable.
Some of these ideas are yet to be evaluated and empirically proven, an undertaking for which we are preparing tests and empirical studies. References can be found at
Fig. 5: 3D focus map along an inner city route